Nitrogen (NH?-N), a crucial water-quality indicator, requires rapid and precise detection for effective pollution control. Conventional chemical colorimetric and titration methods are restricted by high reagent toxicity, low effi-ciency, and susceptibility to interference. Here, a hybrid UV-Vis spectroscopy strategy combining Savitzky-Golay (SG) filtering, Lasso regression, and Elastic Net Regression (ENR) was proposed to achieve rapid, reagent-free detection of NH?-N SG filtering suppresses high-frequency noise while preserving spectral features, Lasso regres-sion identifies seven key wavelengths for data dimensionality reduction, and ENR mitigates multicollinearity via combined L1 and L2 regularization. The SG-Lasso-ENR model shows excellent performance on both training (R2 = 0.9996, RMSE = 0.1271 mg·L?1) and test sets (R2 = 0.9995, RMSE = 0.1101 mg·L?1), with a detection limit of 0.0636 mg·L?1. Its precision (RSD <1%) and accuracy (RE <1%) meet practical requirements. This method eliminates chemical pretreatment, prevents secondary pollution, and offers an efficient, eco-friendly solution for real-time water-quality monitoring, advancing intelligent environmental detection technologies.